214 A.J. Mauri and M. p. Michaels an industry (Barney, 1991). While industrial we design and carry out an empirical test which organization has been primarily concerned with examines this complementarity the similarities among firms, the resource-based view has focused on the differences as the basis to develop sustainable competitive advantage METHODOLOGY (Wernerfelt, 1995 However,empirical studies within industrial Statistical test and measures organization and resource-based view have not The empirical analysis was conducted using the addressed this complementarity. This has been variance components methodology. Unlike partially because of the difficulty in operationaliz. regression techniques using fixed-effects models, ing the theoretical constructs. Researchers within variance components assume a random model industrial organization have relied on proxies of that does not require direct measurement of the structure such as entry barriers, concen- independent variables. Using variance compo- ratios, and industry dummy variables to nents, the unique firm characteristics are modeled strategy and performance. They have as latent factors, captured using individual latent reported a significant effect on r&d and advertis- variables for each firm; while industry attributes ing intensities, as well as performance(see are captured using a common latent variabl Scherer and Ross, 1990, Chs. ll, 16, and 17 for shared by members of the same industry. By a review ) On the other hand, several resource- assuming that the latent factors are selected ran- based studies (Jacobson, 1988; Hansen and Wer- domly from a population of firms and industries, nerfelt, 1989: Powell, 1996) have reported evi- variance components are capable of estimating dence supporting the influence of firm factors on the portion of the total variance derived from performance outcomes despite the difficulties in firm- and industry-specific sources. As rec measuring unobservable firm-specific character- ommended by Searle, Casella and McCulloch istics( Godfrey and Hill, 1995). (1992), the variance component tests were est Even those studies that used more sophisticated mated using the maximum-likelihood method techniques to overcome the measurement prob- The dependent variables for core strategies lems of industry structure and firm-specific were developed for technology and marketing characteristics reported findings that confirmed resources. Several researchers have suggested that he polar perspectives. Both Schmalensee(1985) these intangible resources are a potential source and Rumelt (1991) used variance component of competitive advantage. Dierickx and Cool analysis to study differences in performance (1989) argue that the accumulation process for derived from industry and firm effects. However, generating brand loyalty and technological exper their results are in confict due to methodological tise provides uniqueness and reduces imitation differences. Schmalensee found dominance of because of resource mass efficiencies, resource industry effects as he selected latent variables to erosion, resource interrelationships, time capture industry effects and market share for firm- compression diseconomies and causal ambiguity. level effects; Rumelt found dominance of firm R&d expenditure captures an enterprises endow effects because latent variables were used to cap- ment of unique knowledge possessed by individ ture both the industry and firm-specific effects. uals and teams within organizations( Caves, 1982 In addition, Hill and Deeds(1996)argue that the MacDonald, 1985), and these investments require 3-year time series used by Rumelt is too short periods from 4 to 6 years to provide a return to allow equilibrium to be reached, and Powell ( Cohen and Levin, 1989). Similarly, advertising (1996)questions the validity of the FtC data expenditure captures a firms intangible assets base used. The above discussion brings out the such as brand name and reputation (Stewart, need for more empirical research to test the com- Harris, and Carleton, 1984 ). Because of the high plementarity between the industrial organization uncertainty, high asset specificity and high sunk and resource-based view. In the following section costs associated with R&D and advertising, these expenditures are core strategies financed with equity (Balakrishnan and Fox, 1993), or with 如m时 mo chaure mr galrdgts the mueller9. d r&d and advertisin e 1998 John wiley Sons, Ltd. Strat. Mgm.J,vol.19.211-219(199